Systematic evaluation of lung tumor motion using four-dimensional computed tomography.

BACKGROUND

Respiratory-induced lung tumor motion may decrease robustness and outcome of radiation therapy (RT) if not accounted for. This study provides detailed information on the motion distribution of lung tumors for a group of 126 patients treated with stereotactic body RT.

MATERIAL AND METHODS

Four-dimensional computed tomography scans were reviewed to assess lung tumor motion. The tumor motion was determined by the center of mass shift based on a rigid registration of the breathing phases containing the largest positional differences in the inferior-superior (IS), left-right (LR), and anterior-posterior (AP) directions. The patients were divided into subgroups depending on tumor diameter (φ < 2.0 cm, 2.0 ≤ φ ≤ 5.0 cm, φ > 5.0 cm) and tumor location within the lung (upper, middle, or lower lobe). The observed motion distributions were evaluated for each group separately to assess the dependence on tumor size and location. For each tumor size, the motion pattern in each direction (IS, LR, and AP) was analyzed for every tumor moving >5 mm. Sinusoidal trigonometric functions were fitted to the measured data using the least mean square method to determine which type of function best describes the motion pattern. Tumor volumes between 1.6 and 52.3 cm were evaluated. Mann-Whitney statistical tests were used for statistical analyses.

RESULTS

The mean amplitude for the tumors in this study was 1.5 mm (LR), 2.5 mm (AP), and 6.9 mm (IS) while the maximum amplitude was 11.0 mm (LR), 9.0 mm (AP), and 53.0 mm (IS). In total, 95% of the tumors moved ≤20 mm in the IS direction, ≤3 mm in the LR direction, and ≤6 mm in the AP direction. The observed motion distributions showed no statistically significant correlation with tumor size or location within the lung except for motion in the IS direction, where the mean and maximum amplitudes significantly increased for tumors located in the middle and lower parts of the lung. The motion pattern of a tumor in any direction was best described using a squared trigonometric function of the type [Formula: see text], where A is the maximum amplitude of the motion in the current direction, t is the time of measurement, T is the total time of the breathing cycle and B is a constant used to synchronize the starting point of the breathing cycle.

CONCLUSION

Lung tumor movements were generally larger in the IS direction and the motion amplitude in this direction increased for tumors located in the middle and lower parts of the lungs. Motions in LR or AP showed no such relation. Tumor size was not found to have any correlation with the motion amplitude in any direction. The motion pattern of a lung tumor in any direction is best described with a squared sinusoidal function independently of the tumor size or tumor location.